Improve embedding test coverage

embedding/commentfilter/c_style_filter_test.go

@@ -139,4 +139,31 @@ var _ = Describe("C and C++", func() {
 
 		assertFiltered("sample.cpp", RetainNone, lines, expected)
 	})
+
+	It("should preserve escaped and unterminated quoted literals", func() {
+		lines := []string{
+			`const char* escaped = "quote: \\""; // comment`,
+			`const char* open = "unterminated`,
+			`const char* invalid = R"bad delimiter(text)bad delimiter";`,
+			`const char* missing = R"delimiter`,
+		}
+		expected := []string{
+			`const char* escaped = "quote: \\""; // comment`,
+			`const char* open = "unterminated`,
+			`const char* invalid = R"bad delimiter(text)bad delimiter";`,
+			`const char* missing = R"delimiter`,
+		}
+
+		assertFiltered("sample.cpp", RetainNone, lines, expected)
+	})
+
+	It("should keep an open block comment across lines", func() {
+		lines := []string{
+			"/* open block",
+			"continued",
+			"closed */ int value = 1;",
+		}
+
+		assertFiltered("sample.cpp", RetainBlock, lines, lines)
+	})
 })

embedding/commentfilter/csharp_filter_test.go

@@ -165,4 +165,31 @@ var _ = Describe("C#", func() {
 
 		assertFiltered("Api.cs", RetainNone, lines, expected)
 	})
+
+	It("should preserve C# string variants inside interpolation expressions", func() {
+		cases := [][]string{
+			{`var value = "quote: \\"";`},
+			{`var value = $"{Format("a\\\"b", @"a "" b", """raw""")}";`},
+			{`var value = $"{Format("""raw`},
+			{`var value = $"{Format("text`},
+			{`var value = $"{number:000`},
+			{`var value = $"""Keep {{ and }} {number}""";`},
+			{`var value = @"a "" b";`},
+			{`var value = "unterminated`},
+		}
+
+		for _, lines := range cases {
+			assertFiltered("Api.cs", RetainNone, lines, lines)
+		}
+	})
+
+	It("should keep an open block comment across lines", func() {
+		lines := []string{
+			"/* open block",
+			"continued",
+			"closed */ var value = 1;",
+		}
+
+		assertFiltered("Api.cs", RetainBlock, lines, lines)
+	})
 })

embedding/commentfilter/filter_test.go

@@ -44,6 +44,19 @@ func TestCommentFilter(t *testing.T) {
 }
 
 var _ = Describe("Comment filter", func() {
+	It("should parse supported comment modes", func() {
+		mode, err := ParseMode("")
+		Expect(err).ShouldNot(HaveOccurred())
+		Expect(mode).Should(Equal(RetainAll))
+
+		mode, err = ParseMode(string(RetainBlock))
+		Expect(err).ShouldNot(HaveOccurred())
+		Expect(mode).Should(Equal(RetainBlock))
+
+		_, err = ParseMode("unsupported")
+		Expect(err).Should(HaveOccurred())
+	})
+
 	Describe("unsupported extensions", func() {
 		It("should return unsupported files unchanged", func() {
 			lines := []string{

embedding/commentfilter/java_style_filter_test.go

@@ -30,6 +30,7 @@ import (
 	. "embed-code/embed-code-go/embedding/commentfilter"
 
 	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
 )
 
 var _ = Describe("Java-style languages", func() {
@@ -120,4 +121,15 @@ var _ = Describe("Java-style languages", func() {
 
 		assertFiltered("Api.java", RetainNone, lines, expected)
 	})
+
+	It("should support configured documentation line markers", func() {
+		filter := MarkerCommentFilter{Syntax: CommentMarker{
+			Documentation: DocumentationMarker{Inline: []string{"///"}},
+		}}
+		lines := []string{"/// docs", "code"}
+
+		result := filter.Filter(lines, RetainDocumentation)
+
+		Expect(result).Should(Equal(lines))
+	})
 })

embedding/commentfilter/javascript_filter_test.go

@@ -193,4 +193,31 @@ var _ = Describe("JavaScript and TypeScript", func() {
 
 		assertFiltered("sample.ts", RetainNone, lines, expected)
 	})
+
+	It("should preserve JavaScript regex, string, and interpolation variants", func() {
+		lines := []string{
+			`const text = "escaped \\" // literal"; // comment`,
+			`const classPattern = /[\\/]/gi; // comment`,
+			`/start/.test(value); // comment`,
+			`const open = /unterminated`,
+			`const division = identifier / 2; // comment`,
+			"const template = `escaped \\` text ${value}`; // comment",
+			"const nested = `${{ value: /[}]/g, text: \"}\" }}`; // comment",
+			`const unusual = */ /regex/; // comment`,
+			`*/ /regex/; // comment`,
+		}
+		expected := []string{
+			`const text = "escaped \\" `,
+			`const classPattern = /[\\/]/gi; `,
+			`/start/.test(value); `,
+			`const open = /unterminated`,
+			`const division = identifier / 2; `,
+			"const template = `escaped \\` text ${value}`; ",
+			"const nested = `${{ value: /[}]/g, text: \"}\" }}`; ",
+			`const unusual = */ /regex/; `,
+			`*/ /regex/; `,
+		}
+
+		assertFiltered("sample.ts", RetainNone, lines, expected)
+	})
 })

embedding/commentfilter/kotlin_filter_test.go

@@ -168,4 +168,23 @@ var _ = Describe("Kotlin", func() {
 
 		assertFiltered("Sample.kt", RetainNone, lines, expected)
 	})
+
+	It("should preserve escaped strings, characters, and nested interpolation braces", func() {
+		lines := []string{
+			`val quote = '\'' // comment`,
+			`val text = "escaped \\" // literal" // comment`,
+			`val nested = "${if (ready) { "// literal" } else { value }}" // comment`,
+			`val raw = "${"""raw"""}" // comment`,
+			`val quoted = "${"text"}" // comment`,
+		}
+		expected := []string{
+			`val quote = '\'' `,
+			`val text = "escaped \\" `,
+			`val nested = "${if (ready) { "// literal" } else { value }}" `,
+			`val raw = "${"""raw"""}" `,
+			`val quoted = "${"text"}" `,
+		}
+
+		assertFiltered("Sample.kt", RetainNone, lines, expected)
+	})
 })

embedding/commentfilter/python_filter_test.go

@@ -120,4 +120,42 @@ var _ = Describe("Python", func() {
 
 		assertFiltered("module.py", RetainNone, lines, expected)
 	})
+
+	It("should preserve strings nested inside f-string expressions", func() {
+		cases := []struct {
+			lines    []string
+			expected []string
+		}{
+			{
+				lines:    []string{`value = f"{format('escaped \\' quote')}" # comment`},
+				expected: []string{`value = f"{format('escaped \\' quote')}" # comment`},
+			},
+			{
+				lines: []string{`value = f"{format("""raw`, `text`, `""")}" # comment`},
+				expected: []string{
+					`value = f"{format("""raw`, `text`, `""")}" `,
+				},
+			},
+			{
+				lines:    []string{`value = f"{number:04`},
+				expected: []string{`value = f"{number:04`},
+			},
+			{
+				lines:    []string{`value = f"{format('unterminated`},
+				expected: []string{`value = f"{format('unterminated`},
+			},
+			{
+				lines:    []string{`value = f"{ {'nested': value} }" # comment`},
+				expected: []string{`value = f"{ {'nested': value} }" `},
+			},
+			{
+				lines:    []string{`value = "unterminated`},
+				expected: []string{`value = "unterminated`},
+			},
+		}
+
+		for _, tc := range cases {
+			assertFiltered("module.py", RetainNone, tc.lines, tc.expected)
+		}
+	})
 })

embedding/commentfilter/visual_basic_filter_test.go

@@ -91,4 +91,10 @@ var _ = Describe("Visual Basic", func() {
 
 		assertFiltered("Module.vb", RetainDocumentation, lines, expected)
 	})
+
+	It("should preserve an unterminated quoted string", func() {
+		lines := []string{`Dim value = "unterminated`}
+
+		assertFiltered("Module.vb", RetainNone, lines, lines)
+	})
 })